TABLE 4 
OFF-LINE RADIOMETRIC QUALITY INDEX (RQI) FOR SEVEN BANDS 
(Units are quantum levels. Numbers in parenthesis 
indicate the number of scans that exceeded the 
2 quantum level limit.) 
530 
  
  
  
BANDS 
Area Number of 
Scans ] 2 3 4 5 6 7 
] 32 1.33% 1.60 ^ .79:7,38:.:129055,23^53,95 175 
2 26 1.2142 ,47:0,795 5,60 5;,6]1395,265^1.91* 16) 
3 32 1.38: 7265:^.76^ .3955'185^5:,.22' "3.60^ (32) 
4 32 1.9805:5.,42: 164 158099100717 9550246:089 
5 32 1.23.45 :.77.-.,54: 5.97 .30::1,33^ (10) 
6 32 1.2] .80 .,79 .49 .85 .26 .97 
7 32 1.21.4235 705 933—069 —.24-...54-(23 
8 32 1:26 :.52:5.64: 5.33.70 5.238 ' .87 
9 32 ):3759,265--491 379.245.3499 .92 
  
  
  
LESSONS LEARNED 
Data processing system designers and implementers 
are always faced with some hard realities that can be 
expressed in many ways, but we will generalize as follows: 
1. No matter how well you plan, you won't think of 
everything. 
2. Whatever approach you select, it will require 
at least one major change. 
3. You have to use what is available now, not what 
will be available in a year. 
There are many more, of course, but the point we 
want to make is that the successful implementation of any 
processing system is likely to be an evolutionary process. 
This was especially true in the case of the TIPS. 
In responding to the challenge of building a 
"production" data processor of data from an R&D instrument, 
the performance of which was undefined, an extraordinary 
amount of attention and planning was given to key processes